Sebastian G. wrote:

> nemo_outis wrote:
>
>
> > If you have some argument to show how an unencrypted partition
> > table would permit decrypting the contents of of an encrypted
> > partition, then make it.
>
>
> It doesn't. What it permits is to differ the encrypted disc from
> random data, and it permits knowledge about the partitioning of
> the volume inside the encrypted container.

Only half right. Knowing what type of data might be contained in an
encrypted volume does in general assist in cryptanalysis. Some
forms of this attack are known as "watermarking". Taken to the
extreme it's called a "known plaintext" attack. You have heard
those terms before, haven't you?

And yes, before we start quibbling about the differences so
you can ignore the obvious similarities, those differences exist.
However not as markedly as you may suspect at first jerk. Knowing
that an encrypted volume contains in fact can lead to an actual
known plaintext attack if you're aware of the encrypted volume's
topography (freely published knowledge in this case), and have
knowledge of where certain things will reside within that volume.
Since Windows places certain things in specific areas of a disk,
knowing what's contained inside that encrypted volume enables an
easier collation, and ultimately, the possibility of a successful
attack.

Not that I'm aware of any sort of exploitable known plaintext
weakness in Truecrypt of course. I believe it to be quite secure.
But in general this demonstrates one possible weakness that might
be introduced in a plaintext partition table scenario. And if you
really consider things broadly, it spotlights why OTP is considered
the only truly unbreakable form of encryption. If a ciphertext can
potentially be "anything", it's impossible to even know if you've
successfully decrypted it or not. ;)